A hyperball fuzzy neural network algorithm is proposed for modeling of uncertain, high-dimensional and complex
nonlinear systems based on clustering. Firstly, an improved fuzzy cluster method(FCM) is given to determine the number of fuzzy rules. The one-dimensional membership functions are replaced by the multi-dimensional membership functions. Then, a one-pass algorithm is presented to calculate the centers and parameters of membership functions. The proposed approach can reduce the number of fuzzy rules and simplify the network calculation. Moreover, the fuzzy rules base can be modified online when the input-output data changes. The simulation results show the effectiveness of the proposed approach.